Bulukani M. – Principal Consultant for AI (ADAICO)

At ADAICO, we spend a lot of time exploring how AI can add value to businesses, especially through the broad capabilities of Generative AI agents. This is part of an ongoing series I’m working on. In my latest article, I delved into the potential of Generative AI agents to accelerate the Research Process. During this exploration, I came across ASIC Report 801, which provides a deep dive into the regulatory compliance issues stemming from financial institutions’ failures to adhere to Internal Dispute Resolution (IDR) obligations. The statistics in the report were eye-opening and really highlighted the gravity of the problem for financial services firms. Here are some key stats that stood out:

• Over 4.7 million individual complaints were made to financial firms.
• 2,016 financial firms reported IDR data for the period from 1 July 2023 to 30 June 2024.
• More than $375 million was provided in total monetary remedies to complainants.
• 623,555 complaints resulted in a monetary remedy.
• 5,035 firms declared no complaints, which was higher than expected.

Overall, this paints a picture of systematic compliance and reporting issues. Given how expensive and challenging this problem is, I started to wonder how AI, especially Generative AI, could help address these challenges. I reached out to someone in the industry to gain insights on the underlying issues, and here’s the basic gist of that conversation:

• Humans struggle to design good forms, which has severe and often underestimated impacts on data quality and process outcomes.
• Humans aren’t great at filling out forms, contributing to issues that hinder key aspects of the business process.
• Humans can’t read fast enough to effectively categorise complaints from error-ridden datasets and respond appropriately within strict timelines. This is compounded by the fact that responses require varying levels of complexity and correct contextualisation.

This was enlightening; what seemed like a simple problem of understanding and responding to complaints quickly is actually a tedious and challenging task for human operators. Fortunately, the issues my contact identified can be managed efficiently with the help of AI. With careful planning, organisations can clean and understand their datasets, enabling them to respond better, faster, and cheaper. As I mentioned in my Generative AI series, successful use of Generative AI agents relies heavily on clear, well-defined business processes, but I want to focus on potential ways Generative AI can be effectively used to tackle problems like this:

• Automated Complaint Classification: Generative AI can automatically classify incoming complaints by type, enabling quicker responses and appropriate escalation paths. This ensures that complaints are managed in line with regulatory requirements, reducing the risk of non-compliance and enhancing overall efficiency.
• Intelligent Response Generation: For standard complaints, templated responses can be designed, allowing firms to acknowledge complaints promptly. This not only improves the customer experience but also ensures compliance with mandatory disclosure requirements.
• Seamless Escalation: Complaints that require human intervention can be seamlessly escalated with all relevant information, expediting resolution and ensuring compliance with RG271 timelines. This addresses the need for timely resolution, as many firms struggle to meet regulatory timeframes.
• Automated Data Capture and Validation: Generative AI can be integrated into data management and processing systems to streamline data entry processes, minimising the risk of human error. This ensures that all relevant complaint data is accurately recorded, providing a reliable dataset for analysis and reporting. Implementing validation checks during data entry enhances the accuracy and completeness of the data collected, supporting firms in meeting ASIC reporting requirements.

I hope you found this journey as insightful and enlightening as I have. I’m constantly surprised by how many expensive, challenging problems can be addressed with existing AI technology, thanks to the amazing capabilities of Generative AI. Please let me know your thoughts—I’m really interested in investigating and sharing how Real AI can solve Real problems for real people!